Lamin R. Mansaray

Mapping Rice Fields in Urban Shanghai, Southeast China, Using Sentinel-1A and Landsat 8 Datasets

Sentinel-1A and Landsat 8 images have been combined in this study to map rice fields in urban Shanghai, southeast China, during the 2015 growing season. Rice grown in paddies in this area is characterized by wide inter-field variability in addition to being fragmented by other landuses. Improving rice classification accuracy requires the use of multi-source and multi-temporal high resolution data for operational purposes. In this regard, we first exploited the temporal backscatter of rice fields and background land-cover types at the vertical transmitted and vertical received (VV) and vertical transmitted and horizontal received (VH) polarizations of Sentinel-1A. We observed that the temporal backscatter of rice increased sharply at the early stages of growth, as opposed to the relatively uniform temporal backscatter of the other land-cover classes. However, the increase in rice backscatter is more sustained at the VH polarization, and two-class separability measures further indicated the superiority of VH over VV in discriminating rice fields. We have therefore combined the temporal VH images of Sentinel-1A with the normalized difference vegetation index (NDVI) and the modified normalized difference water index (MNDWI) derived from a single-date cloud-free Landsat 8 image. The integration of these optical indices with temporal backscatter eliminated all commission errors in the Rice class and increased overall accuracy by 5.3%, demonstrating the complimentary role of optical indices to microwave data in mapping rice fields in subtropical and urban landscapes such as Shanghai.

Dynamic Mapping of Rice Growth Parameters Using HJ-1 CCD Time Series Data

The high temporal resolution (4-day) charge-coupled device (CCD) cameras onboard small environment and disaster monitoring and forecasting satellites (HJ-1A/B) with 30 m spatial resolution and large swath (700 km) have substantially increased the availability of regional clear sky optical remote sensing data. For the application of dynamic mapping of rice growth parameters, leaf area index (LAI) and aboveground biomass (AGB) were considered as plant growth indicators. The HJ-1 CCD-derived vegetation indices (VIs) showed robust relationships with rice growth parameters. Cumulative VIs showed strong performance for the estimation of total dry AGB. The cross-validation coefficient of determination ( R C V 2 ) was increased by using two machine learning methods, i.e., a back propagation neural network (BPNN) and a support vector machine (SVM) compared with traditional regression equations of LAI retrieval. The LAI inversion accuracy was further improved by dividing the rice growth period into before and after heading stages. This study demonstrated that continuous rice growth monitoring over time and space at field level can be implemented effectively with HJ-1 CCD 10-day composite data using a combination of proper VIs and regression models.

Estimation and Mapping of Winter Oilseed Rape LAI from High Spatial Resolution Satellite Data Based on a Hybrid Method

Leaf area index (LAI) is a key input in models describing biosphere processes and has widely been used in monitoring crop growth and in yield estimation. In this study, a hybrid inversion method is developed to estimate LAI values of winter oilseed rape during growth using high spatial resolution optical satellite data covering a test site located in southeast China. Based on PROSAIL (coupling of PROSPECT and SAIL) simulation datasets, nine vegetation indices (VIs) were analyzed to identify the optimal independent variables for estimating LAI values. The optimal VIs were selected using curve fitting methods and the random forest algorithm. Hybrid inversion models were then built to determine the relationships between optimal simulated VIs and LAI values (generated by the PROSAIL model) using modeling methods, including curve fitting, k-nearest neighbor (kNN), and random forest regression (RFR). Finally, the mapping and estimation of winter oilseed rape LAI using reflectance obtained from Pleiades-1A, WorldView-3, SPOT-6, and WorldView-2 were implemented using the inversion method and the LAI estimation accuracy was validated using ground-measured datasets acquired during the 2014–2015 growing season. Our study indicates that based on the estimation results derived from different datasets, RFR is the optimal modeling algorithm amidst curve fitting and kNN with R2 > 0.954 and RMSE <0.218. Using the optimal VIs, the remote sensing-based mapping of winter oilseed rape LAI yielded an accuracy of R2 = 0.520 and RMSE = 0.923 (RRMSE = 93.7%). These results have demonstrated the potential operational applicability of the hybrid method proposed in this study for the mapping and retrieval of winter oilseed rape LAI values at field scales using multi-source and high spatial resolution optical remote sensing datasets. Details provided by this high resolution mapping cannot be easily discerned at coarser mapping scales and over larger spatial extents that usually employ lower resolution satellite images. Our study therefore has significant implications for field crop monitoring at local scales, providing relevant data for agronomic practices and precision agriculture.

Mapping deforestation and urban expansion in Freetown, Sierra Leone, from pre- to post-war economic recovery.

Freetown, the capital of Sierra Leone has experienced vast land-cover changes over the past three decades. In Sierra Leone, however, availability of updated land-cover data is still a problem even for environmental managers. This study was therefore, conducted to provide up-to-date land-cover data for Freetown. Multi-temporal Landsat data at 1986, 2001, and 2015 were obtained, and a maximum likelihood supervised classification was employed. Eight land-cover classes or categories were recognized as follows: water, wetland, built-up, dense forest, sparse forest, grassland, barren, and mangrove. Land-cover changes were mapped via post-classification change detection. The persistence, gain, and loss of each land-cover class, and selected land conversions were also quantified. An overall classification accuracy of 87.3xa0% and a Kappa statistic of 0.85 were obtained for the 2015 map. From 1986 to 2015, water, built-up, grassland, and barren had net gains, whereas forests, wetlands, and mangrove had net loses. Conversion analyses among forests, grassland, and built-up show that built-up had targeted grassland and avoided forests. This study also revealed that, the overall land-cover change at 2001–2015 was higher (28.5xa0%) than that recorded at 1986–2001 (20.9xa0%). This is attributable to the population increase in Freetown and the high economic growth and infrastructural development recorded countrywide after the civil war. In view of the rapid land-cover change and its associated environmental impacts, this study recommends the enactment of policies that would strike a balance between urbanization and environmental sustainability in Freetown.

Evaluating the potential of temporal Sentinel-1A data for paddy rice discrimination at local scales

ABSTRACT A mapping algorithm is proposed in this letter on the application of Sentinel-1A data in discriminating paddy rice from other land-cover categories at local scales. The study region is Chongming Island located in the Shanghai metropolitan area, southeast China. We have acquired five temporal images of the new Sentinel-1A satellite in interferometric wide swath (IW) mode, covering a critical period of the 2015 paddy rice growing season in Chongming Island. Temporal backscatter at vertical transmitted and horizontal received (VH) polarization of Sentinel-1A was exploited and we observed that from early June to mid September, temporal backscatter profiles of the classes water/pond, built/urban, trees/forest and others were relatively stable. On the other hand, paddy rice exhibited a marked change in temporal backscatter coefficients, increasing steeply from flooding/planting to tillering/booting, and decreasing slightly at heading. Backscatter profiles also differ between paddy rice fields of different flooding/planting periods. These observed temporal microwave (radar) backscatter dynamics were employed in a decision tree mapping algorithm to discriminate paddy rice from other land-cover classes. An overall classification accuracy and Kappa statistic of 88.3% and 0.85 were recorded, respectively, which demonstrates the operational applicability of temporal Sentinel-1A data in paddy rice discrimination at local or district scales.

The Influence of Different Spatial Resolutions on the Retrieval Accuracy of Sea Surface Wind Speed With C-2PO Models Using Full Polarization C-Band SAR

This paper presents a comparison strategy for investigating the influence of spatial resolutions on sea surface wind speed retrieval accuracy with cross-polarized synthetic aperture radar images. First, for wind speeds retrieved from vertical transmitting-vertical receiving (VV)-polarized images, the optimal geophysical C-band model (CMOD) function was selected among four CMOD functions. Second, the most suitable C-band cross-polarized ocean (C-2PO) model was selected between two C-2POs for the VH-polarized image data set. Then, the VH-wind speeds retrieved by the selected C-2PO were compared with the VV-polarized sea surface wind speeds retrieved using the optimal CMOD, which served as a reference, at different spatial resolutions. Results show that the VH-polarized wind speed retrieval accuracy increases rapidly with the decrease in spatial resolutions from 100 to 1000 m, with a drop in root-mean-square error of 42%. However, the improvement in wind speed retrieval accuracy levels off with spatial resolutions decreasing from 1000 to 5000 m. This demonstrates that the pixel spacing of 1 km may be the compromising choice for the tradeoff between the spatial resolution and wind speed retrieval accuracy with cross-polarized images obtained from RADASAT-2 fine quad-polarization mode.

Assessing Global Ocean Wind Energy Resources Using Multiple Satellite Data

Wind energy, as a vital renewable energy source, also plays a significant role in reducing carbon emissions and mitigating climate change. It is therefore of utmost necessity to evaluate ocean wind energy resources for electricity generation and environmental management. Ocean wind distribution around the globe can be obtained from satellite observations to compensate for limited in situ measurements. However, previous studies have largely ignored uncertainties in ocean wind energy resources assessment with multiple satellite data. It is against this background that the current study compares mean wind speeds (MWS) and wind power densities (WPD) retrieved from scatterometers (QuikSCAT, ASCAT) and radiometers (WindSAT) and their different combinations with National Data Buoy Center (NDBC) buoy measurements at heights of 10 m and 100 m (wind turbine hub height) above sea level. Our results show an improvement in the accuracy of wind resources estimation with the use of multiple satellite observations. This has implications for the acquisition of reliable data on ocean wind energy in support of management policies.

A universal estimation model of fractional vegetation cover for different crops based on time series digital photographs

Abstract The green fractional vegetation cover (FVC) is an important parameter in monitoring crop growth and predicting aboveground biomass. In this study, we monitored crop growth with digital cameras installed at four automatic weather observation stations in different parts of China, from 2010 to 2016. With each station having a particular type of crop, nine color vegetation indices were calculated from the acquired time series digital photographs to arrive at an FVC estimation model applicable to sugarcane, maize, cotton and paddy rice. For individual crop types, our results show that the Excess Green (ExG) is the optimal color vegetation index for the estimation of sugarcane FVC, the Normalized Difference Index (NDI) is the optimal color vegetation index for the estimation of maize FVC, and the Vegetative (VEG) color vegetation index is optimal for the FVC estimation of cotton and paddy rice. However, owing to its higher coefficient of determination (R 2 ), and lower root mean square error (RMSE) and mean absolute error (MAE) of 0.9504, 0.0721 and 0.0545, respectively, the Color Index of Vegetation Extraction (CIVE) is found more universally applicable for FVC estimation of the four crop types under investigation. The CIVE index has therefore been proposed in this study to be optimal for FVC estimation in sugarcane, maize, cotton and paddy rice mixed cropping agro-systems which are especially common in small and highly fragmented agricultural landscapes such as those in urban and peri-urban areas.

Correction: Wang, J., et al. Dynamic Mapping of Rice Growth Parameters Using HJ-1 CCD Time Series Data. Remote Sens. 2016, 8, 931

All stages EVI2 E 0.358 10.210 cu EVI2 Q 0.923 18.247 B 0.362 10.193 B 0.918 18.452 S 0.444 9.968 S 0.921 32.613 NDVI E 0.275 10.798 cu NDVI Q 0.929 17.621B 0.334 10.460 B 0.922 17.964 S 0.467 10.185S 0.927 32.092 Before heading EVI2 E 0.831 6.074 cu EVI2 Q 0.909 25.317 B 0.926 6.152B 0.901 26.932 S 0.900 6.776 S 0.884 45.126 NDVI P 0.644 8.960 cu NDVI Q 0.922 23.496B 0.615 9.023 B 0.902 25.187 S 0.629 10.363 S 0.920 40.714 After heading EVI2 E 0.421 8.036 cu EVI2 Q 0.481 15.067 B 0.474 8.019 B 0.474 15.998 S 0.416 8.205 S 0.571 14.862 NDVI E 0.496 7.607 cu NDVI Q 0.516 14.632 B 0.610 8.630 B 0.426 13.207 S 0.657 7.076 S 0.573 14.587